Innovation and Industry Development: Lessons from the British Cotton Textile Industry During the U.S. Civil War

This dissertation uses the large shock to the British cotton textile industry in the 19th century, caused by the U.S. Civil War (1861-1865), in order to address three long-running questions about technological progress and industry development. The cotton textile industry was a large and important sector in the British economy during the 19th century. The industry was entirely dependent on imported raw cotton, most of which came from the U.S. South prior to the Civil War. The onset of the war sharply reduced the supply of Southern cotton to the British market, causing a severe downturn in the industry. In response to the shock, cotton textile producers turned to other sources of supply, chiefly India, but also Egypt, Brazil and others, to help meet their raw cotton needs. But cotton from these alternative suppliers, and India in particular, was very different from the U.S. cotton that British producers were used to spinning. As a result, British cotton textile producers were faced with a number of new challenges. The first two chapters of this dissertation describe how the cotton textile industry developed new technology in order to deal with these challenges, and what this response can tell us about the process of innovation. Chapter three then investigates the impact of the recession on other industries in the British economy. This historical setting has a number of features which makes it a particularly good setting for investigating technological progress. First, the U.S. Civil War caused a shock which was both large and exogenous. The size of the shock ensures that the response will be large enough to clearly observe, while the exogenous nature of the shock means that it can be used as a natural experiment in order to uncover causal relationships. Second, the impact of the U.S. Civil War was largely industry-specific; while the impact of the war on the cotton textile industry was severe, most other sectors of the British economy were not directly impacted. This includes other textile industries based on wool, linen, and silk, which do not show any ill effects during the war. One advantage of this feature is that it allows me to control for other time-varying factors by comparing the cotton textile industry to these other similar industries. I will also be able to uncover evidence of inter-industry connections, since other industries will be affected primarily through their relationship with the cotton textile industry. Another feature of this shock is that, despite the magnitude, there was virtually no government intervention in the affected markets. This unique feature was due to the particularly strong free-market ideology that was dominant in Britain during this period. The first chapter investigates the theory of directed technical change. The leading theory of directed technical change, developed by Acemoglu (2002), offers two main predictions. First, when inputs are sufficiently substitutable, a change in relative input supplies will generate technical change directed towards the inputs which become more abundant. Second, if technical change is strongly directed towards the more abundant inputs, the relative price of these inputs will increase -- the strong induced-bias hypothesis. The chapter provides the first empirical test of these predictions. I extend the theory to a setting in which input quantities are endogenous and affected by international transport cost shocks, such as that caused by the war. Using detailed new patent data, I show that there was a burst of cotton textile innovation in Britain during the war directed towards taking advantage of one input -- Indian cotton -- which became relatively more abundant. Next, I show that the relative price of Indian cotton first declined and then rebounded, consistent with the strong induced-bias hypothesis. These results provide support for the theory. My extended model also predicts that technical change directed towards the more abundant input will be magnified by a higher elasticity of input supply. This may explain why inventors chose to focus on innovations for Indian cotton, rather than Brazilian or Egyptian cotton, since I find evidence that the elasticity of supply was higher for Indian cotton. In the second chapter, I look at whether the stock of available knowledge about a particular type of technology can influence the rate of innovation in that technology. The answer to this question has significant implications for how we think about technological progress and economic growth. This chapter provides a theory which describes how the stock of knowledge can influence the innovation rate, which I call path dependence in innovation. The theory suggests that path dependence in innovation may occur at multiple levels of aggregation, such as specific types of technologies within an industry. This motivates an empirical exercise in which I search for path dependence at multiple technology levels. I introduce an empirical methodology that addresses two potential sources of bias in generating evidence of path dependence in innovation by using a temporary observable shock to innovation rates. My results provide no evidence of path dependence in innovation for cotton textile technologies. However, I do find suggestive evidence of path dependence in innovation for specific subsets of technologies within the cotton textile industry. This illustrates the importance of looking at multiple levels of aggregation when studying path dependence in innovation. Chapter 3 provides causal evidence that inter-industry connections can influence the geographic location of economic activity. To do so, it compares the impact of the shock caused by the U.S. Civil War on towns in Lancashire County, the center of Britain's cotton textile industry, to towns in neighboring Yorkshire County, where wool textiles dominated. The results suggest that the shock reduced employment and employment growth in industries related to the cotton textile industry, in towns that were more severely impacted by the shock, relative to less affected towns. The impact still appears over two decades after the end of the U.S. Civil War. This suggests that temporary shocks, acting through inter-industry connections, can have long-term impacts on the distribution of industrial activity across locations. Each of the three chapters are entirely self-contained, so that a reader interested in only one of these topics need focus on only one chapter. As a result, each chapter contains an overview of the features of the empirical setting which are relevant for that chapter. There are significant overlaps between these descriptions. Also, chapters one and two are largely based on the same British patent data, though each chapter uses some parts of the data which are not used in the other